Download Conjunctive and disjunctive verb forms

Conjunctive and disjunctive verb forms
D. Langa (Edoardo Mondlane University) & T. Taraldsen (University of Trosmø)
1. The basic contrast
In certain tenses, the verb has two morpho-syntactically distinct forms with the same
TAM-semantics, a conjoint (CJ) form and a disjoint (DJ) form, on many Bantu
languages. The CJ (verb form) must be followed by a (VP-internal) complement or an
(VP-internal) adverb, while a DJ need not be followed by anything. In other tenses. the
same form appears in both environments. We will call such forms neutral.
We will consider CJ and DJ forms as morphologically distinct even in the absence of
segmental morphology whenever the two forms have distinct tone patterns that cannot
fully be accounted for by general tone rules.
2. Two approaches to the CJ vs. DJ contrast
We know of two main theoretical interpretations of the empirical facts. On the one hand,
researchers following Hyman & Watters (1984) take the CJ vs. DJ contrast to reflect
focus placement. According to Hyman & Watters, for example, the DJ forms appear if
and only if the truth value of the sentence is in focus (“assertive focus”), i.e. the
occurrence of DJ morphology is directly comparable to the appearance of the emphatic
did in English sentences like He DID eat the apple.
A different view is taken by van der Spuy (1993) and others who claim that DJ
morphology simply signals that the verb is VP-final. (Notice that “VP” in this context
must stand for a constituent big enough to contain both the verb and (focused) adverbs.)
Occasionally, it is suggested that focus-placement is the determinant factor in some
languages, while other languages require an analysis along the lines of van der Spuy; cf.
Buell & Riedel (2008). But we will suggest that some characteristic properties of the CJ
vs. DJ contrast are too stable cross-linguistically for this proposal to be plausible. Since
those properties seem to speak in favor of an analysis akin to Hyman & Watters’s
proposal, we are therefore led to favor approaches associating the CJ vs. DJ alternation
with focus placement in all Bantu languages, even though there are well-known
counterexamples that are taken to argue against. One of these is mentioned in section 6,
where we suggest that it can be circumvented on a more elaborate analysis of focusing.
3. Properties favoring Hyman & Watters (1984)
The following are the properties that seem to be characteristic of the CJ/DJ-alternation in
all Bantu languages:
A. The DJ includes the CJ
Riedel (2009) notes that a CJ can be turned into a DJ by the addition of a morpheme to its
left, reflecting Hyman & Watters claim that “the focus variants are derived from the [focus] forms by the addition of something”. But it does not appear to be the case that a
DJ is ever turned into a CJ by the addition of an overt morpheme. This seems easier to
understand under a focus-based analysis than in the analyses emanating from van der
Spuy (1993) which would have to explain why there is an overt marker when the verb is
VP-final, but none when it is not VP-final, rather than the other way around.
B. Sensitivity to TAM properties
The CJ/DJ alternation is found only with certain tense/aspect combinations, typically the
present (non-progressive) and the perfect. Other TAM categories are typically neutral, e.g.
the future or the progressive forms. While Hyman & Watters’s account may provide the
beginnings of an understanding of this (see section 6), analyses adhering to van der
Spuy’s view must take the limitation of the CJ/DJ-contrast to certain TAM-categories to
be accidental. This will count as an argument against those analyses to the extent that the
sensitivity to TAM-properties can be shown not to be random. We think that it is not, but
to demonstrate this, we will need a precise characterization of the semantics of TAMs
across the different Bantu languages.
C. Sensitivity to negation
There is generally no contrast between CJ and DJ forms with identical TAM-properties
in negative sentences. Only the CJ forms appear, e.g. the Nguni –ya- never appears under
negation. Again, it seems that analyses incorporating van der Spuy’s hypothesis have
nothing to offer, while Hyman & Watters can take the negation itself to be able to license
“assertive focus”. (However, the present perfect shows a DJ/CJ-alternation even under
negation in Zulu according to Buell (xx).)
D. Limitation to certain clause-types
The alternation between CJ and DJ forms with the same TAM properties is only found in
indicative sentences. There are no purely DJ-forming morphemes like the Nguni –yawith participial or subjunctive verbs, although real TAMs may occur. This too seems
unexpected on van der Spuy’s proposal, but Hyman & Watters can account for it by
showing that certain clause types are incompatible with assertive focus.
We would like to emphasize that claim that the CJ/DJ-alternation has these properties in
all Bantu languages, must obviously be tested on the basis of data from more languages
than we have been able to look at. But it is significant that they are shared both by
Aghem for which a focus-based analysis seems plausible and Zulu, which van der Spuy’s
claim was based on.
In the next two section, we look at other empirical issue that need to be settled.
4. Objects in and out of VP
Saying that DJ morphology reflects assertive focus, makes predictions about the position
of the object of a verb in its DJ guise only to the extent that one also assumes that
assertive focus excludes associating the object with focus, and that non-focused objects
must be “dislocated”. In principle, there might be languages in which a DJ form allows
the object to remain inside the VP. This obviously contrasts with analyses based on van
der Spuy’s proposal. Therefore, it is important to determine whether there are languages
in which DJ morphology clearly is compatible with having an object in the VP. If there
are such languages, the van der Spuy proposal simply cannot be the basis for a unified
analysis of DJ vs. CJ contrasts across the Bantu languages. By contrast, finding languages
in which the object must be evicted from the VP when the verb is a DJ form, doesn’t
contradict the focus-based analysis.
To this end, one needs to examine the position of the object with respect to VP-external
elements in sentences with DJ verb forms. If the object must be dislocated when the verb
appears in its DJ form, it would be expected to follow VP-external elements. But there is
some evidence that this is not universally the case. For example, Hyman & Watters’
examples (1) and (2) from Aghem show that the object of a DJ verb precedes nε “today”
just like the object of a CJ verb, and that nε only precedes the object, when nε itself is
focused and the verb is in its CJ form.
Prosodic evidence may also bear on this issue. For example, Kraal (2009) provides
examples from Makonde where a DJ form is followed by a constituent with which it
seems to form a prosodic domain exactly as the CJ form does.
5. Types of focus and the syntactic feature +Focus
In Luganda, a noun can lose its initial vowel only if it is focused. In Aghem, the class
marker of a noun is prefixed to the noun when the noun is focused, but is a suffix when
the noun is out of focus. In both languages, the “in-focus form” of an object cannot cooccur with DJ-morphology on the verb. (Notice that Hyman & Watters’ example (1)b
indicates that the out-of-focus form can still be VP-internal.) On Hyman & Watters’
approach this is accounted for by assuming that only one constituent per clause can be
focused. But this seems to presuppose that assertive focus and object focus etc. at some
level count as the same thing, i.e. assigning assertive focus is just one of several
equivalent ways of providing a (unique) focus. On this view, certain clauses (essentially
indicative clauses) come with a syntactic feature +Focus (possibly located in the upper
region of the clause) which must be associated with some constituent in the clause, but
doesn’t care which. A DJ form appears just in case the (extended) VP is to be associated
with +Focus and there is no TAM which can mediate this association.
If this is correct, the clause types in which DJ forms do not appear must lack the feature
+Focus. But if focusing an object or an adverb also involves association with +Focus, this
leads to the expectation that the distinct in-focus forms of object nouns should not appear
in these clause types either with the important qualification that the object of a participial
verb in certain constructions might be accessed by the matrix +Focus, while assertive
focus on the participial VP would be blocked by the matrix verb. It is therefore important
to verify if this correlation holds in all Bantu languages where nouns arguably have
distinct in-focus forms. If it does, we have another argument against simple
implementations of van der Spuy’s idea, which provides no reason to expect that in-focus
forms of nouns and DJ forms of verbs should fail to appear in exactly the same clausetypes.
We should add that contrastive focus cannot be dependent on +F within this scheme,
since Hyman & Watters show that a contrastive focus marker may be associated with an
out-of-focus noun in Aghem.
Finally we look at one type of counterexample to the focus-based account of CJ/DJalternations.
6. Auxiliary focus vs. VP-focus
Buell (xx) shows that even in a main clause the verb always appears in the CJ form in
Zulu when it is followed by kahle “well”. even though kahle itself may not be focused.
This is hard to understand on Hyman & Watters approach where DJ morphology is
associated with assertive focus, since He DID sing well is perfectly fine.
To make Buell’s observation consistent with the analysis outlined in section 5., we need
to say that the appearance of kahle within the VP allows the VP itself to be focused
without the intermediary of assertive focus, and that assertive focus in Zulu is a last resort
mechanism invoked only when it is the only way of associating a constituent with +F.
The first of these two claims embodies a claim about the way sets of alternatives are
formed, assuming that focus is based on the existence of a set of alternatives, as in Rooth
(1992): We suggest that a set of alternatives to a focused VP cannot be constructed by
changing the verb or its arguments, but can be based on low modifiers like kahle (in a
way that doesn’t require kahle itself to be contrastively focused).
When the VP itself cannot be the basis for the construction of a set of alternatives,
alternatives can only be formed with respect to TAM properties introduced at higher
nodes. This is where auxiliary focus kicks in. In tenses that come with a TAM morpheme
that expresses TAM values that can be used for the formation of alternatives, we then get
neutral forms. If there is no such TAM morpheme, a DJ morpheme must be merged
(assertive focus).
It is implicit in this that not all TAMs provide properties that allow the construction of
alternatives. Departing from Hyman & Watters account, we will suggest that some TAMs
are so deeply embedded that their TAM values are already fixed and cannot be played
with to form alternatives at the point where they are seen by the relatively high +Focus
head. For example, the same auxiliary can mean both future “will” and deontic “must” in
Kîîtharaka, but has a DJ form only when it means “must”, and we take this difference to
reflect the fact that this auxiliary occurs low when it means “must”, but high when it
means “will”, in a way consonant with the general thrust of Ramchand’s (2012) account
of modals.
7. The project
This survey of analyses and the data bearing on them is meant to highlight the need to fill
in certain gaps in our understanding of what the patterns really are. We propose to use
AfrAnaph resources to this end while being aware of it that eliciting reliable judgments
about fine-grained TAM semantics is a tall order.
References:
Buell, L. & K. Riedel (2008), The conjoint/disjoint alternation in Sambaa, TiN-dag handout
Hyman, L. & J.R. Watters, Auxiliary focus, Studies in African linguistics 15.3
Kraal, P. (2009), Makonde, in G. Dimmendaal (ed), Coding participant marking:
Construction types in 10 African languages, Vol. 110, John Benjamins
Ramchand, G. (2012), Indexical vs. anaphoric modals, ms., CASTL, Tromsø
Riedel, K. (2009), The syntax of object marking in Sambaa, LOT dissertation series
Rooth, M. (1992), A theory of focus interpretation, Natural Language Semantics 1.1
van der Spuy, A. (1993), Dislocated noun phrases in Nguni, Lingua 90, 335-355